Open Source Data Backup, or: How to Sleep Better at Night (OSCON 2005)

Open Source Data Backup, or:
How To Sleep Better At Night
Fran Fabrizio
Senior Systems Administrator
Dept. of Computer and Information Sciences
University of Alabama at Birmingham
O’Reilly Open Source Convention, August 1-5, 2005

Open Source Data Backup (OSCON 2005), Slide 2
Talk Overview
• Introduction to Amanda
– What it is, how it thinks
• Amanda In Action
– Real world examples
• Configuring Amanda
– Quick look at the config files
• Introduction to Bacula
– What it is, how it thinks
• Bacula vs. Amanda
– Major differences
• For More Info

Introduction To Amanda
What is Amanda?
How does it work?
What is its philosophy?

What is Amanda?
• Cross-Platform
• Scalable
• Automated
• Flexible
• Robust
Amanda is the Advanced Maryland Automatic
Network Disk Archiver. It has the following
features:

What is Amanda? (Cont)
• A set of CLI utilities written in C
• Its own protocols on top of TCP and UDP
• Client and server components that partner to
stream and store data for backup and recovery

Holding
Disk
Tape
Drive
OS X Client
Solaris Client
Linux Client Windows
Client
Amanda
Amanda Server
amandad
Samba
amandad
amandad
Sample Topology

Typical Sequence of Events
Amanda Server Amanda Client
amdump
planner
dumper(s)
driver
taper
amandad
amandad
launches
passes schedule
spawns
writes
to disk
writes to
taper
flush to tape
requests estimate
returns estimate
requests backup
returns image
(many clients serviced in parallel)

How Does It Work?
• gtar and dump
• Uses standard backup levels
• Manages tapes
• Balances resources
• Supports compression and encryption
• Degrades gracefully

Some Amanda Terminology
• Dump cycle
– How often do you want a full backup?
• Disklist Entry / DLE / Target
– Something (partition, filesystem) you want to backup

Amanda's Philosophy
• “You tell me how often you want a full backup,
and I'll worry about everything else”
• At least one full backup of each DLE per cycle
• You don't get to say when full backups happen
• Sounds scary, but usually sufficient

Real World Examples – CIS @ UAB
• Before....
(And this isn't even the whole dirty truth)

• So we bought...
Great! Only
one problem...

• This hardware is very expensive
• I work for a public university...
• ....in Alabama ;-)
• We had no more money
• Amanda saved the day
• Usage at CIS
– 43 filesystems/partitions on 14 Linux, Solaris and
Windows clients
– ~ 60GB of data per night
– Fully automated, requiring about 2 hours of attention so
far this year

Real World Examples - Others
• Other Examples
– Much Larger
• One user wrote about a 700GB nightly dump
– Much Smaller
• one client systems
– Vtape setups
• virtual tapes on disk
• periodically burn to DVD

Typical Amanda Daily Operation
• At 2:30pm, am check utility runs via cron job and
informs me via email if there are any problems
• At 2:00am, am dum p utility runs to kick off the
Amanda backup process
• Sometime in the morning, amdump sends an email
summarizing last night's activity.

Sample amdump Email Output
These dumps were to tape CIS-004.
The next tape Amanda expects to use is: CIS-005.
STATISTICS:
Total Full Daily
-------- -------- --------
Estimate Time (hrs:min) 1:00
Run Time (hrs:min) 2:57
Dump Time (hrs:min) 3:24 2:02 1:22
Output Size (meg) 20957.1 17678.4 3278.8
Original Size (meg) 38473.8 31408.7 7065.1
Avg Compressed Size (%) 54.5 56.3 46.4 (level:#disks ...)
Filesystems Dumped 41 11 30 (1:28 3:1 4:1)
Avg Dump Rate (k/s) 1753.8 2474.3 682.5
Tape Time (hrs:min) 0:21 0:15 0:05
Tape Size (meg) 20957.2 17678.4 3278.8
Tape Used (%) 4.3 3.6 0.7 (level:#disks ...)
Filesystems Taped 41 11 30 (1:28 3:1 4:1)
Avg Tp Write Rate (k/s) 17361.9 19931.6 10242.2
USAGE BY TAPE:
Label Time Size % Nb
CIS-004 0:21 20957.2 4.3 41

Sample amdump Email Output
(Continued)NOTES:
planner: Full dump of virginia:/hc promoted from 4 days ahead.
planner: Full dump of florida:/var promoted from 4 days ahead.
planner: Full dump of alabama:/etc promoted from 4 days ahead.
planner: Full dump of florida:/etc promoted from 1 day ahead.
planner: Full dump of alabama:/home promoted from 1 day ahead.
planner: Full dump of georgia:/home promoted from 4 days ahead.
planner: Full dump of illinois:/home promoted from 4 days ahead.
planner: Full dump of newyork:/home promoted from 4 days ahead.
planner: Full dump of newjersey:/ promoted from 4 days ahead.
taper: tape CIS-004 kb 21462176 fm 41 [OK]
DUMP SUMMARY:
DUMPER STATS TAPER STATS
HOSTNAME DISK L ORIG-KB OUT-KB COMP% MMM:SS KB/s MMM:SS KB/s
-------------------------- --------------------------------- ------------
missouri.hpcl.c -xport/home 1 24671 13205 53.5 3:42 59.6 0:018848.1
missouri.hpcl.c /export/opt 1 1055 49 4.6 0:44 1.1 0:01 42.8
missouri.hpcl.c /var/mail 1 1538303 828724 53.9 13:171040.4 1:299345.3
nevada.cis.ua / 4 2271760 881675 38.8 5:092850.4 1:0813033.1
ohio.cis.u /home 1 1540 247 16.0 0:01 401.8 0:04 69.7
maine.cis. /home 1 82640 59078 71.5 0:144300.2 0:087671.2
florida.cis.u /etc 0 5200 1613 31.0 0:03 625.5 0:011186.9
florida.cis.u /he 1 8780 614 7.0 0:26 23.4 0:01 512.8
florida.cis.u /hf 1 16440 2066 12.6 2:02 16.9 0:021249.4
etc....
(brought to you by Amanda version 2.4.4p2)

Amanda's planner In Action
planner clientA
“Please estimate level 0, level 1 and level 2 backups for /home.”
Let's assume that
last night Amanda
performed a level 1
backup of clientA's
/home DLE...
clientA returns the info
“Will promoting this to a level 0 (full) dump lead to
more balance over the dump cycle?”
“Will we save a significant amount of tape space by
going to a level 2 incremental instead of level 1?”
Schedules
Level 0 dump
Schedules
Level 2 dump
Schedules level 1 dump
Yes
Yes
No
No

More about planner
• Guarantees one full dump per cycle
• If the tape is too small for the run's data, it will
delay some of the dumps in the least disruptive
way
• Looks at past dumps to determine optimal
balancing behavior
• Tries to stay as close to level 0 as possible to
reduce need to use multiple tapes for restore

Restoring One Or A Few Files
[root@alabam a /tm p]# amrecover -C CIS -s
amanda.cis.uab.edu -t amanda.cis.uab.edu -d
/dev/nst0
AMRECOVER Version 2.4.4p1.Contacting server on
am anda.cis.uab.edu ...
220 keep AMANDA index server (2.4.4p2) ready.
200 Access OK
Setting restore date to today (2004-06-28)
200 Working date set to 2004-06-28.
Scanning /dum ps/am anda...
20040622:found Am anda directory.
200 Config set to CIS.
200 Dum p host set to alabam a.cis.uab.edu.
am recover> setdisk /var/spool/mail
200 Disk set to /var/spool/m ail.
am recover> setdate 2004-06-25
200 Working date set to 2004-06-25.
Problem: User 'bryant' requests that you restore his INBOX to its state on 2004-06-25.
Solution: Use the amrecover utility

Restoring One Or A Few Files (Con't)
am recover> ls
[...]
2004-06-25 brockhw
2004-06-25 brownta
2004-06-25 bryant
2004-06-25 byrdv
[...]
am recover> add bryant
Added /bryant
am recover> extract
Extracting files using tape drive /dev/nst0 on host am anda.cis.uab.edu.
The following tapes are needed:CIS-024
Restoring files into directory /tm p
Continue [?/Y/n]? Y
Extracting files using tape drive /dev/nst0 on host am anda.cis.uab.edu.
Load tape CIS-024 now
Continue [?/Y/n/s/t]? Y
./bryant
am recover> quit
200 Good bye.
[root@alabam a /tm p]# ls -l bryant
-rw------- 1 bryant disk 14533946 Jun 24 19:10 bryant

Restoring An Entire Backup Target
Problem: The disk holding /etc on a critical server has failed
Solution: Use the amrestore utility.
[root@am anda testrestore]# amadmin CIS info alabama.cis.uab.edu '/etc$'
Current info for alabam a.cis.uab.edu /etc:
Stats:dum p rates (kps),Full: 481.0,320.0,350.0
Increm ental: 19.0, 13.0, 11.0
com pressed size,Full: 20.6%,20.6%,20.6%
Increm ental: 5.4%, 5.4%, 5.4%
Dum ps:lev datestm p tape file origK com pK secs
0 20040623 CIS-022 23 18690 3849 8
1 20040628 CIS-027 10 700 38 0
[root@am anda testrestore]# su - amanda
-bash-2.05b$ am tape CIS labelCIS-022
am tape:scanning for tape with labelCIS-022
am tape:slot 26:date 20040623 labelCIS-022 (exact labelm atch)
am tape:labelCIS-022 is now loaded.
-bash-2.05b$ exit

Restoring An Entire Backup Target (Con't)
[root@am anda testrestore]# amrestore -p /dev/nst0 alabama.cis.uab.edu /etc >
etc.0.tar
am restore: 0:skipping start oftape:date 20040623 labelCIS-022
am restore: 1:skipping am anda.cis.uab.edu.__texas_dfs_hom e_undergrad.private.20040623.1
am restore: 2:skipping florida.cis.uab.edu._etc.20040623.1
am restore: 3:skipping alabam a.cis.uab.edu._root.20040623.1
[...]
amrestore: 21: skipping missouri.cis.uab.edu._.20040623.0
amrestore: 22: skipping oregon.cis.uab.edu._home.20040623.1
amrestore: 23: restoring alabama.cis.uab.edu._etc.20040623.0
amrestore: 24: reached end of information
[root@am anda testrestore]# ls -l
total18716
-rw-r--r-- 1 root root 19138560 Jun 28 14:40 etc.0.tar

Other Ways To Restore Files
• Amanda not available?
root@am anda testrestore]# mt -f /dev/nst0 rewind
[root@am anda testrestore]# dd if=/dev/nst0 bs=32k count=1
AMANDA:TAPESTART DATE 20040623 TAPE CIS-022
1+0 records in
1+0 records out
[root@am anda testrestore]# mt -f /dev/nst0 fsf 01
[root@am anda testrestore]# dd if=/dev/nst0 bs=32k count=1
AMANDA:FILE 20040623 am anda.cis.uab.edu
//texas/dfs/hom e/undergrad.public lev 1 com p .gz program /usr/bin/sm bclient
To restore,position tape at start offile and run:
dd if=<tape> bs=32k skip=1 |/bin/gzip -dc |usr/bin/sm bclient -f...-
1+0 records in
1+0 records out

When Things Go Wrong
• Data backup is a complex interaction between a
lot of players, and things -will- go wrong...
– Hosts will be down, or away (laptops)
– Tapes will go bad
– You'll change a password and then forget to tell Amanda
you changed it
– Your holding disk might be too small
• Amanda tries to be proactive by running am check
and giving you time to fix problems

Sample amcheck Problem Report
• Host down
Amanda Tape Server Host Check
-----------------------------
Holding disk /dumps/amanda: 60967688 KB disk space available, that's plenty
amcheck-server: slot 5: date 20040520 label CIS-018 (exact label match)
NOTE: skipping tape-writable test
Tape CIS-018 label ok
Server check took 175.627 seconds
Amanda Backup Client Hosts Check
--------------------------------
WARNING: vermont.cis.uab.edu: selfcheck request timed out. Host down?
Client check: 13 hosts checked in 30.210 seconds, 1 problem found
(brought to you by Amanda 2.4.4p2)

More amcheck Sample Messages
NOTE: index dir
/
usr/local/etc/amanda/CIS/index/amanda.cis.uab.edu/__texas_dfs_home_undergr
ad.private: does not exist
ERROR: georgia.cis.uab.edu: [access as amanda not allowed from
amanda@amanda] amandahostsauth failed
ERROR: virginia.cis.uab.edu: [dir /etc needs 64KB, only has 5KB
WARNING: holding disk /dumps/amanda: only 50254708 KB free (52428800 KB
requested)
amcheck-server: slot 19: rewinding tape: No medium found
amcheck-server: slot 19: date 20040330 label CIS-027 (active tape)
amcheck-server: fatal slot 20: slot 20 move failed
ERROR: label CIS-016 or new tape not found in rack
(expecting tape CIS-016 or a new tape)
amcheck-server: could not get changer info: could not read result from
"/usr/local/libexec/chg-scsi"
WARNING: skipping tape test because amdump or amflush seem to be running
WARNING: if they are not, you must run amcleanup

amcheck Is The Best Thing Since Sliced Bread
• Cron this script to email you every day before you
leave the office
• Run this script manually any time you touch a
config file
• Run this script manually any time you add or
change a client
• It will save you hours of troubleshooting
• One of the best features of Amanda

amcheck Can't Do Everything
• am check cannot catch problems that arise after it
runs
• am check doesn't check for everything
• Amanda's other utilities are really good at telling
you why they could not do their job

Revisiting The amdump Email Report
• Very rarely is the email report as uneventful as
was presented earlier. It typically will include
something like this:
These dumps were to tape CIS-004.
The next tape Amanda expects to use is: CIS-005.
FAILURE AND STRANGE DUMP SUMMARY:
alabama.cis.u /usr lev 1 STRANGE
STATISTICS:
Total Full Daily
-------- -------- --------
Estimate Time (hrs:min) 1:00
Run Time (hrs:min) 2:57
etc....

Revisiting The amdump Email Report (Con't)
• And then later on, it will explain itself....
FAILED AND STRANGE DUMP DETAILS:
/-- alabama.cis.u /usr lev 1 STRANGE
sendbackup: start [alabama.cis.uab.edu:/usr level 1]
sendbackup: info BACKUP=/bin/gtar
sendbackup: info RECOVER_CMD=/bin/gzip -dc |/bin/gtar -f... -
sendbackup: info COMPRESS_SUFFIX=.gz
sendbackup: info end
? gtar: ./local/majordomo-1.94.5/log/mdlog: file changed as we
read it
| Total bytes written: 207923200 (198MB, 3.1MB/s)
sendbackup: size 203050
sendbackup: end
--------
Amanda is telling us a file was in use as it tried to
grab it. This is usually harmless and constitutes
99% of STRANGE results.

Revisiting The amdump Email Report
(Con't)
• Sometimes, there are more serious failures
FAILURE AND STRANGE DUMP SUMMARY:
amanda.cis.u //texas/dfs/simnetxpcd lev 0 FAILED [disk
//texas/dfs/simnetxpcd, all estimate failed]
amanda.cis.u //texas/dfs/classfiles lev 0 FAILED [disk
//texas/dfs/classfiles, all estimate failed]
amanda.cis.u //texas/dfs/officefiles lev 0 FAILED [disk
//texas/dfs/officefiles, all estimate failed]
amanda.cis.u //texas/dfs/scripts lev 0 FAILED [disk //texas/dfs/scripts,
all estimate failed]
amanda.cis.u //texas/dfs/home/graduate.private lev 0 FAILED [disk
//texas/dfs/home/graduate.private, all estimate failed]
amanda.cis.u //texas/dfs/home/graduate.public lev 0 FAILED [disk
//texas/dfs/home/graduate.public, all estimate failed]
amanda.cis.u //texas/dfs/home/undergrad.private lev 0 FAILED [disk
//texas/dfs/home/undergrad.private, all estimate failed]
This was because I changed the password to the SAMBA
share that Amanda was using to back up a Windows server.
The email clued me in and the problem was resolved
quickly.

Handling Tape Failures
• Amanda keeps going and stores as much as
possible in the holding disk
• You can then use am flush to flush data to tape
-bash-2.05b$ amflush CIS
Scanning /dum ps/am anda...
Multiple Am anda directories,please pick one by letter:
A.20040516
B.20040517
Select directories to flush [A..B]:[ALL]
Today is:20040517
Flushing dum ps in 20040516,20040517 using tape changer "chg-scsi".
Expecting tape CIS-015 or a new tape. (The last dum ps were to tape CIS-014)
Are you sure you want to do this [yN]? y
Running in background,you can log offnow.
You'llget m ailwhen am flush is finished.

Amanda Prerequisites
• A server that is mostly idle during the times that
you want to do your backups
• Enough disk space for a suitable holding disk
• GNU tar
• Samba (for Windows clients)
• A large capacity tape drive (typically)
• GNUplot

Configuration Roadmap
• Set up server
– Create amanda user and assign to group with permission
to use the tape and changer devices (e.g. 'disk' on Linux)
– Gather info on tape and changer devices - mt and mtx
are handy scripts to have here
– Open ports (10080,10082,10083) and set up services
(amanda, amandaidx, amidxtape)
– Configure amanda.conf & changer (e.g. chg-scsi.conf)
• Set up clients
– Create amanda user
– Config ports, services, files, directories
– Allow access (typically .amandahosts)

Things To Think About Before
Proceeding
• What should my cycle be?
– Once per night? Every three nights?
• If your cycle seems convoluted, try using two:
– One for daily backups, once per night, one week cycle
– One for archives, always full dumps, run manually when
you need it
• Find the right balance for the cycle length
– Short cycles eat up lots of resources doing full dumps
– Long cycles can be a pain to restore from, you might
need 4 tapes for example
• How many tapes to use?
– How far back into the past do you want to go?

amanda.conf
• Amanda's main configuration file
• Many options
• Well documented
• Cannot begin to cover everything here, definitely
read the documentation first!
• This is where you define your cycle's parameters
(length, number of runs, number of tapes, etc...)

amanda.conf - dumptypes
• Different rules for each DLE
define dum ptype root-tar {
global
program "GNUTAR"
com m ent "root partitions dum ped with tar"
com press none
index yes
exclude list "/usr/local/lib/am anda/exclude.gtar"
priority low
}
define dum ptype com p-high {
global
com m ent "very im portant partitions on fast m achines"
com press client best
priority high
}

disklist
• Defining your DLEs / targets
# File form at is:
#
# hostnam e diskdev dum ptype [spindle [interface]]
# the tape server itself
am anda.cis.uab.edu / com p-root-tar-exclude-holdingdisk
# the directory server
newjersey.cis.uab.edu / com p-root-tar
# the file server
virginia.cis.uab.edu /m z com p-user-tar
virginia.cis.uab.edu /root com p-root-tar
virginia.cis.uab.edu /usr com p-user-tar
virginia.cis.uab.edu /hc com p-root-tar

Configuring Tape Changers
• Amanda has a generic interface to tape changers
• Tape changer configuration is stored in chg-
scsi.conf
• chg-scsi.conf is one of many changer scripts that
come with Amanda
• This is where you tell Amanda how many drives
you have, which tapes go to which drives, barcode
support, etc...

Tape Drive Configuration Heads-Up
• There are many choices of changer configuration
scripts (chg-scsi, chg-multi, chg-mtx, chg-manual,
chg-disk, etc...)
• You may have to use one even if you don't have
an actual changer (chg-manual, chg-disk)
• Many options, many chances for confusion
• Please read docs/TAPE.CHANGERS and other
sources of information (listed at end of talk)

Configuring Your Tape Collection
• Use amlabel to label new tapes
– am labelCIS CIS-000 slot 0
• Then use amtape to build the tapelist
– am tape CIS update
am tape:scanning all30 slots in tape-changer rack:
slot 26:date 20040623 labelCIS-022
etc...

Special Case Configurations
• Windows Clients
– Use Samba to backup Windows clients
– Configure the Amanda server or other Unix client with
Samba shares
– Then in the disklist point to that server and share
•am anda.cis.uab.edu //texas/dfs/officefiles com p-user-tar

Special Case Configurations
• Windows Clients
– Use Samba to backup Windows clients
– Configure the Amanda server or other Unix client with
Samba shares
– Then in the disklist point to that server and share
•am anda.cis.uab.edu //texas/dfs/officefiles com p-user-tar
• Firewalls
– --with-tcpportrange=40000,40030 ( something > 1024)
– --with-udpportrange=920,940 (something < 1024)
– or, iptables has amanda support you can enable

Introduction To Bacula
What is Bacula?
How is it different?

What is Bacula?
• “The Network Backup Tool for Linux, Unix, Mac
and Windows.”
• Another open-source project that aims to provide
a robust network-based, multiplatform backup
solution
• Newer than Amanda (started in 2000)
“It comes in the night and sucks the essence from your computers.”
- Kern Sibbald

Bacula Features
• Modular, scalable components
• Its own protocols on top of TCP and UDP
• Client and server components that partner to
stream and store data for backup and recovery
• Clean component separation - all communication
between them goes over the network
• Threaded rather than multiple processes
• Excellent documentation

How is Bacula Different than Amanda?
• It can support multiple volumes
• You may find it easier to setup
• There are both command line and GUI
configuration tools available
• Scheduler gives you more control over what jobs
run at which times
• It will reuse a tape on multiple nights until it is
full
• Support for automated restores from bare metal
• Native support for Windows (no Samba/NFS)
• SQL database support

Sample Bacula Topology
Director
Daemon
Database
Server
File Daemon
Windows
File Daemon
Linux
File Daemon
Unix
File Daemon
OS X
Storage
Daemon
Tape
Device
Storage
Daemon
Tape
Changer
Storage
Daemon
Disk
Device
Admin
Console

Component Roles
• The Director manages all scheduling and job
creation. It is via an administration interface
talking to the Director that the backup
administrator controls the backup process.
• The Storage Daemon is responsible for writing
data out to disk/tape/changer
• The Database keeps the catalog of what has been
backed up and where
• The File Daemon streams the data to be backed up
from the client to the Storage Daemon

Prerequisites
• Bacula currently works with SQLite, MySQL and
PostgreSQL
• GNU C++ 2.95 or higher to compile
• Other software may be necessary depending on
configuration. If burning DVDs, you need the
dvd+rw-tools. If using the GUI console, you need
recent GNOME and GTK+ libs.

Configuration Overview
• Bacula is configured via a series of config files
– bacula-dir.conf, bacula-fd.conf, bacula-sd.conf,
console.conf
• Examples coming here

Storage Organization
• Bacula organizes tapes (volumes) into pools. It
will use one volume up until it is full, and then
move on to the next. You can give more control if
you want to use a new tape each night
• Similar to Amanda, each Bacula volume get a
unique label and added to the pool. However,
unlike Amanda, there is no set rotation. Bacula
will use one until full, and then go look for
another, etc...
• Can use multiple pools to ensure that a new tape is
used each day. e.q. Setup a Monday pool, a
Tuesday pool, etc...

Pre and Post Job Scripts
• Bacula's File Daemon has the ability to run a
script before and after a job
• This can be used to shutdown a database in order
to take a safe backup of it, for example
• Use the bacula-fd.conf directives “Run Before
Job” and “Run After Job”

Take Home Points...
• Amanda is very robust, highly scalable, almost
infinitely configurable, and can very likely handle
your data backup situation
• If you find Amanda limiting or do not agree with
the scheduling philosophy, Bacula may be for
you. Momentum -may- be headed to Bacula.
• Once you get either of these configured, you can
trust it, and move on with your life. They work,
are battle-tested and just as reliable, if not more
so, than the expensive commercial products.
• Test your backup and recovery system and
strategy early and often!!

For More Information
• http://www.amanda.org/
• Documentation
– README, INSTALL, docs/* (esp. FAQ and
TAPE.CHANGERS), example/*
– The AMANDA section in O'Reilly's “Unix Backup and
Recovery”
– man pages
• http://www.bacula.org/
– Documentation - Tutorial, Quick Start, User Guide
• The user communities are wonderful
– Sign up for amanda-users@amanda.org at
http://www.amanda.org/support/mailinglists.php
– Sign up for bacula-users@lists.sourceforge.net at
http://lists.sourceforge.net/lists/listinfo/bacula-users

My Information
• My email is fran@cis.uab.edu
• This presentation is available at
http://www.cis.uab.edu/fran/
• More Amanda and Bacula information can be
found at the above URL, including:
– A document detailing every step of my Amanda
configuration, along with complete sample amanda.conf
and chg-scsi.conf
– Notes on configuring Samba to backup Windows shares
– Sample Bacula configuration files for each component
– Notes on strategies for setting up volumes and pools for
Bacula

My Information
• My email is fran@cis.uab.edu
• This presentation is available at
http://www.cis.uab.edu/fran/
• Much more Amanda information can be found at
the above URL, including:
– A document detailing every step of my configuration
– A complete sample amanda.conf and chg-scsi.conf
– Notes on configuring Samba to backup Windows shares
The End - Thank You!

Open Source Data Backup, or: How to Sleep Better at Night (OSCON 2005)

Open Source Data Backup, or: How to Sleep Better at Night (OSCON 2005)

Recommended

More Related Content

What's hot (20)

Similar to Open Source Data Backup, or: How to Sleep Better at Night (OSCON 2005) (20)

Recently uploaded (20)

Open Source Data Backup, or: How to Sleep Better at Night (OSCON 2005)